henhaobio
Sign in

Frontiers in ImmunologyImmunologyPeer-reviewed journal article

replicateFest: An R Package and Shiny App for Analysis of T Cell Receptor Repertoire Data from the Functional Expansion

Ludmila DanilovaiD, Alexander V. FavoroviD, Kellie Nicole SmithiD, Leslie CopeiD

PDFDOIPreprintCodeData
In-depth full-text digest available · Jump to digest ↓

Abstract

MotivationThe Functional Expansion of Specific T cell (FEST)-based assays combine short-term peptide stimulation with TCR sequencing to identify clonotypes that expand in response to specific antigens. These approaches have proven invaluable for detecting neoantigen-specific T cell responses, guiding vaccine development, and assessing checkpoint blockade efficacy. However, variability introduced by biological and technical replicates poses challenges for reproducibility and interpretation, and existing computational tools do not address replicate-level analysis in these assays. ResultsWe developed replicateFest, a computational framework implemented as an R package and Shiny web application, to analyze FEST-based TCR-seq data with and without replicates. replicateFest applies Fishers exact test for non-replicate datasets and negative binomial modeling for replicate experiments, returning adjusted p-values and odds ratios to identify clonotypes significantly expanded in antigen-stimulated conditions. The framework distinguishes FEST-expanded clonotypes (relative to a no-antigen control) and FEST-positive clonotypes (expanded compared to all other conditions). Validation using synthetic datasets confirmed accurate detection of antigen-specific clonotypes. Application to published HIV-1 epitope stimulation data reproduced original findings and demonstrated replicateFests utility for reproducibility assessment and quality control. Availability and ImplementationreplicateFest is freely available under the Apache-2.0 license as an R package at https://github.com/OncologyQS/replicateFest and as an interactive Shiny application at http://www.stat-apps.onc.jhmi.edu/FEST/.

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 3 (13 Jul 2026) · CC BY 4.0

replicateFest uses negative binomial models for replicated and Fisher's exact tests for unreplicated FEST TCR-seq to call antigen-expanded clonotypes, recovering all nine published HIV-1 responders.

FEST (Functional Expansion of Specific T cells) assays combine brief peptide stimulation with TCR sequencing to find antigen-responsive clonotypes, yet biological and technical replicate variability hampers reproducible calls. Existing tools such as ALICE, TCRNET, GLIPH2, tcrdist3 and TCRscape detect sequence convergence or clustering but do not handle replicates, a gap replicateFest fills while retaining the earlier unreplicated workflow.

Key results

  1. ShownReanalysis of a published replicated HIV-1 Gag/Nef epitope-variant stimulation study (Chen et al.) recovered every clonotype the original authors reported as expanded, with small differences in which clones were flagged cross-reactive.

    After dropping samples with fewer than 5,000 productive reads (79 samples; 25 peptide conditions plus no-peptide reference), FDR 0.01 and OR 5 thresholds called all nine reported clones (three Gag, six Nef) FEST-expanded; four of five reported cross-reactive clones recovered, plus one cross-reactive clone not reported by Chen et al. · 2.3 Analysis of HIV-1 data; 3.2 Application to HIV-1 data; Tables

  2. SuggestedSimulations seeded from empirical HIV-1 count distributions indicate that detection power under default thresholds depends jointly on a clone's baseline frequency and its fold expansion, setting practical sensitivity limits.

    10 simulated datasets, each with 90 spiked clones in a 3 × 3 design (baseline at 50th/75th/95th percentiles; 2×, 4×, 8× expansion; three replicates), FDR < 0.05 and OR > 1: near-zero detection at 2×; 0.42 ± 0.10 (L1) to 0.92 ± 0.10 (L3) at 4×; ≥ 0.98 at 8× for all frequencies, no variability at L3. · 2.4.1 Data Generation; 2.4.2 Application of replicateFest; 3.3 Application to the simulated data; Figure legends

  3. ShownClonotypes are called FEST-expanded (significant vs the no-peptide reference, odds ratio above threshold) or, more strictly, FEST-positive (expanded vs every other parallel condition), via per-clone negative binomial models for replicates or Fisher's exact tests otherwise.

    Negative binomial coefficients are read as log odds ratios relative to the no-peptide control, fit jointly across conditions; p-values use Benjamini-Hochberg FDR (default threshold < 0.05); both clone sets export to Excel and FEST-positive clones are drawn as a heatmap; R functions runExperiment and runExperimentFisher. · 2.1.2 Estimating Clone Abundance; 2.1.3 Filtering for Significant and Unique…; 2.1.4 Implementation and availability

  4. SuggestedOn packaged synthetic repertoires with one strongly expanded clone per stimulated condition, replicateFest identified both FEST-expanded and FEST-positive clonotypes in designs with and without replicates.

    Random CDR3 sequences (12 to 16 amino acids; 36 to 45 nucleotides), background counts around 10, one clone per stimulated sample expanded about 100 times; five peptide conditions plus one unexpanded condition; outcome reported qualitatively, without detection metrics. · 2.2 In silico dataset generation; 3.1 In silico experimental data

Methods, data and reproducibility

Imported ImmunoSEQ/VDJtools tables are restricted to productive clonotypes and collapsed by amino acid sequence; per-clone negative binomial models (replicates) or pairwise Fisher's exact tests (no replicates) yield Benjamini-Hochberg-adjusted p-values and odds ratios versus a no-peptide control.

  • Published replicated TCR Vβ CDR3 sequencing of CD8+ T cells stimulated with 25 HIV-1 Gag and Nef epitope-variant peptides (Chen et al.) · human (inferred: T cells responding to autologous HIV-1 epitopes) · n = 79 samples after QC (25 peptide conditions plus reference) · ImmuneACCESS project chan-2020-fi
  • Background count matrix of 351 non-expanded clones (at least 50 total reads) from Chen et al., 10 peptide conditions in triplicate plus reference · human (inferred: T cells responding to autologous HIV-1 epitopes) · n = 33 samples; 351 clones
  • Simulated datasets: negative binomial counts for 90 spiked expanded clones (3 × 3 frequency by fold-expansion design) appended to the background matrix · n = 10 simulated datasets of 90 clones each
  • Synthetic in silico TCR repertoires with random CDR3 sequences, with and without replicates, shipped as package test data · n = five peptide conditions plus one unexpanded condition
Methods & tools
replicateFest (R package and Shiny app), Negative binomial regression, Fisher's exact test, Benjamini-Hochberg FDR correction, Adaptive Biotechnologies ImmunoSEQ export format, VDJtools, Adaptive ImmuneACCESS Database, Loess fit and Akima spline interpolation (simulation dispersion), FEST/MANAFEST peptide-stimulation assay, TCR Vβ CDR3 sequencing, R (v4.6.0)
Software & environment
R package requires R ≥ 4.0; HIV-1 reanalysis performed in R v4.6.0 with replicateFest v0.1.0 (also used for simulations); Apache-2.0 license.
Reproducibility gaps
HIV-1 results deferred to Supplementary Table 1 and synthetic data generation details deferred to the R package; no dependency versions, lockfile or container specified.
Limitations (6)
  • Sensitivity simulations found 2-fold expansions essentially undetectable at all starting frequencies, so modest antigen-driven expansions, especially in rare clones, may be indistinguishable from background count variability. · stated by the authors
  • Detection at 4-fold expansion was strongly frequency dependent and the most variable across simulation runs (0.42 ± 0.10 at L1), which the authors attribute to stochastic sampling of TCR-seq counts. · stated by the authors
  • Cross-reactivity calls on the HIV-1 data did not fully match Chen et al. (one reported cross-reactive clone missed, one additional clone flagged), which the authors ascribe to differing expansion-calling methods. · stated by the authors
  • Single-cell TCR-seq input, paired α/β chain analysis and automated replicate-concordance thresholds are not yet supported and are listed as future development; the current workflow targets TCR Vβ CDR3 data. · stated by the authors
  • No specificity or false-discovery assessment is reported: simulations measure only recovery of spiked clones, and no head-to-head comparison with the Chen et al. filtering or other tools is quantified on the same data. · noted in this summary
  • Real-data validation relies on one published HIV-1 dataset with nine reference positives, analyzed at stricter thresholds (FDR 0.01, OR 5) than the defaults, so default-threshold behavior on real repertoires is untested. · noted in this summary

Read this if you run FEST, MANAFEST, or other peptide-stimulation TCR-seq assays with replicates and need a statistical, reproducible way to call antigen-expanded clonotypes.

Citations
0
FWCI
0.00
Field percentile
top 30%

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISCeglia '21Pogorelyy '19Gielis '19Plessy '15Milighetti '26Vujovic '21
64% semantic matchImmunology
V. Ceglia, E. J. Kelley, A. S. Boyle+3 more·bioRxiv·May 21, 2021